looking for rotary optical encoder disc

Fotofab may have something standard.

John

If you search the web you will find a free program for creating images of rotary encoder discs. Can't remember what it is called or where but it was some kind of homebrew robotics site so you will probably find other useful info about home brewing encoders and discs.

You could have the disk exposed onto lith film in a print shop (you'd have to glue the film onto a tranparent disc afterwards), or have it laser-cut from thin metal. Or have it milled from PCB material in one run with the next PCB you make...

robert

If he's worried about bearing play, kitchen table methods won't help.

HP or whatever they are called this week do a most comprehensive range of disks- what kind of speed is the OP working at if he's getting edge separation problems? They won't go down to 50cpr, as this would make the disk prohibitively small- the sensors are positioned 90 optical degrees apart, where the size of the optical degree is governed by the slot pitch.

50cpr- that's 200 states per rev, or 1.8 degrees. I can't see bearing play affecting this, unless the shaft fits in the hole like the proverbial prick in a top hat.

I suspect that a good look at the electronics will save you much grief.

Paul Burke

Etched pcb material is relatively transparent to infra-red but copper is not. My homemade bobbin winder just used a 10 'slot' etched pcb for the turns counter.

--
Tony Williams.

Le Fri, 17 Aug 2007 10:15:33 +0100, Tony Williams a écrit:

Relying on FR4 transparency won't help jitter, and I think it's a pb to colin. But it's very possible to keep all the copper and strategically place a bunch of holes, not plated to limit reflexion pbs. Using 0.8mm thickness may slightly improve things too.

--
Thanks,
Fred.

I have a pilar drill and a soldering iron, oh and a dremel too... its amasing what you can do acuratly if you have enough inginuity. (even if i cant spell well)

I have a buisiness card made from one of those chemical milling companies its realy cool.

thats quite a good idea actually.

as I said I am at the moment using a heds6100/9100 set. 2000cpr but as you cortectly say they dont go down to 50cpr. i think its avago these days (have a go ?)

im using them upto 10krpm and they perform quite well at well over their spec of 100khz.

I have a 40cpr plastic encoder wich came off an old dot matrix printer, shame I havnt got 2., im also using it for one of the other axis of rotation. I could use it to make a spray paint image onto something.

the heds 6100 have an ID of upto about 6mm, I had to bore them out to 12mm I was amased I did it carefully enough not to scratch the film disc.

if I put another heds 9100 sensor on the oposite side it would require centering the disc to a very small tolerance or will likely change by more than 90' electrical phase or as the shaft rotates.

this is a possibility I was considering but would require more electronics to use pulses from oposite sides and cope with them not being in quadrature. such as QEI absolute position tracking logic, ... im hoping to just xor them.

I might take the plunge and add an fpga where I can just feed all the signals in from 4 encoders, but this solution seemed atractivly simple.

ah yes well the electronics does quite a lot too, it spins at 10krpm and is averaged by a dspic33 over several hours, so thats a lot of points.

its compared to another identical encoder to give an average difference of position to a truly ridiculous resolution. eqv time is sub ps.

ive made other posts about this project were it is pointed out how futile this is, but nevertheless I have acheived sub ps resolution, but im trying to remove some more of the noise.

as long as the pulses from the encoders either side stay in quadrature doesnt matter about the absolute positional acuracy as long as it doesnt change.

im painstakingly trying to get the 1M 12mmdia shaft straight to within 1um. but it tends to go back to its original shape after an hour :(

Colin =^.^=

thanks ive had a look before, and ive tried some graphics programs but it seems dificult to get them to output something usable.

I also asked in a post somewhere for same sort of thing, id apreciate it if you could remember something to reference it by.

Colin =^.^=

ive used a reflective IR opto device elsewhere and put black crepe paper pcb tape on a white plastic disc, but this was only 1 cpr, worked adequatly, exept the tape didnt like to stick to the plastic so well.

im not sure how close the lines could be for the reflective device. Id like to make the disc larger, maybe use a 5" disc platter but id have to modify things a fair bit.

Colin =^.^=

anything here?

Martin

yes thanks thats super :)

however I was just thinking id go for a plastic geer wheel and reflective ir sensor.

I played around with some gear wheels and a sensor I have, seems the plastic type rather than the metal type give a better signal, and with the sensor on the outside of the wheel.

the advantage is that the gear can be withdrawn without having to move the sensor, and is very tolerant to axial alignment.

I can get 48teeth 50mm acetal geer with 8m bore, I gues I can bore this out to 12mm somehow, maybe with a soldering iron ! although I need to ensure very good radial acuracy.

I wonder if those disks from usdigital will work in reflective mode ? I need a sensor on each side rather than 2 next to each other.

I also played around with some sheet tin metal and some lines drawn with a felt tip pen the sensor picked that up quite well, so a printed image shld work OK too.

the gear seems easiest to try as its only a day away from a local supplier.

thanks Colin =^.^=

If you want to try laser printing onto overhead transparency film (at least for a prototype), Postscript is a fairly easy language to write a code generator for. Just look up the applicable primitives and write a program in your favorite language that writes the commands to a .ps file.

I did do this at one point, but needed a higher resolution than producable by a printer... the results turned out more like a moire (?) pattern -undersampling of the geometry at the printer resolution. (Actually I made an 8" diameter disk figuring I'd later modify to generate only the diamter I actually intended to cut out and use... but within the middle two inches it was not useable at the number of CPR I coded).

See if you have a Coin-Op distributor in your town. Some of the encoder wheels used in the large trak-balls are pretty close to what you are talking about. You might see if they have repair parts for rebuilding them. See :

Jim

wow thanks for that, thats realy cool ive got it to draw a simple one, but my inkjet printer is realy ancient.

Ive been meaning to get a test print of one of these latest epson injets, you can send them an image and get a test print back :)

however the resolution of the drawing seems rather low at 1:1 it wont do the inkjet resolution of >1000dpi much justice so I need to scale it up, but ive no idea how to make sure the image wil come out the right size, I gues I just need to make the slots long enough and then trim as necessary.

thanks Colin =^.^=

I have never used it in anger and yes the results seem a bit 'jaggy'. I played with it briefly and it seems to be limited to creating an image the size of the white screen rectangle it is drawn in.

Looks like the best you can do is scale to fill that square then somehow scale back down with the printer driver or print to an image file and edit that.

- Lith film itself, including the "black" areas, is pretty transparent to infrared. You can use it to make a cheesy sort of IR filter, but for an encoder wheel, you'll have to find something else.

Heh. I remember, way back when, maintaining an old Anorad CNC machine that used visible light in its encoders. It was controlled by an DEC LSI-11 of about 9 cubic feet running FORTRAN, a machine which today would be considered an inferior pocket calculator.

I have exactly what he needs, except that there are two sensors on the same side, slightly offset by degrees to actually double the resolution of the reader electronics.

I will post a picture in alt.binaries.schematics.electronic soon to show it.

It has a much higher res than he mentioned as there are about 200 slots around the disc. Almost the exact dimensions he refers to, however, and the clamping mechanism as well. Sits on a 3/8" shaft IIRC.

If he wants it, I'll perhaps put it up on ebay so he can grab it.

Said picture has been posted.

I think the idea of putting two sensors 180 degrees apart is not needed.

If there is that much damned slop in your shaft you are going to have some serious errors anyway you "look" at it.